SQL学习笔记

一.数据库概述

（1）DBMS(数据库管理系统)和数据库,平时谈到"数据库"两种含义:MSSQLSever,Oracle等某种DBMS;存放一堆数据表的一个分类(Catalog)

     不同牌子的DBMS有自己的不同特点
     主键有两种选用策略:业务主键和逻辑主键.
     业务主键是使用有业务意义的字段做主键,比如身份证号,银行账号.
     逻辑主键是使用没有任何业务意义的字段做主键.因为很难保证业务主键不会重复,推荐用逻辑主键
（2）表间关联,外键(ForeignKey)
（3）SQLSever的管理

二.数据库基础

（1）数据类型
     bit位,相当于bool类型,0,1
     char(10) 字符 长度10 不足用空格填充
     int整型
     bigint 超大整型
     nvarchar(MAX) 最大,无限大,可能有中文信息,日文   
     varchar() 无中文信息  不会用空格填充  var(variable)
     datatime 日期
     --注释内容

三.SQL语句表操作

（1）ALTER 修改表
     关键字,表名,变量名对大小写是不敏感的

 CREATE TABLE TABLE_NAME(ID int NOT NULL,NAME nvarchar(50),AGE int NULL)
  INSERT INTO TableName(Column1, Column2, ..., ColumnN) VALUES(Column1Value, Column2Value, ..., ColumnNValue)
  DELETE FROM TableName WHERE whereCondition
  UPDATE TableName SET Column1 = Column1Value, Column2=Column2Value, ..., ColumnN=ColumnNValue WHERE whereCondition
  SELECT * FROM TableName (WHERE whereCondition)

 SQL主要分DDL(数据定义语言)和DML(数据操作语言)两类,Create Table,Drop Table,等属于DDL, select,insert,update,delete,属于DML

（2）主键选择
     常用主键类型,int,bigint,uniqueidentifier(又称GUID,UUID)
     guid算法使用mac地址,地址,纳秒级时间,芯片id码等算出来,每次guid永远不会重复
     .net中生成guid的方法,Guid.NewGuid(),返回Guid类型   SQL中newid(),比如select newid()
     (*)int自增字段的优点:占用空间小,无需开发人员干预,易读;缺点: 效率低,数据导入导出的时候很痛苦
     (*)Guid的优点:效率高,数据导入导出方便; 缺点:占用空间太大,不容易读懂(慢慢成为主流)

（3）数据插入
     可以给字段设置默认值,如果guid类型主键的默认设为newid(),就会自动生成,很少这么干
     中文前面加N
     insert into Person2(Id,Name,Age)values(newid(),'tom',30)

（4）数据更新
     update Person1 set NickName=N'青年人' where Age>=20
     <>不等于
（5）表删除
     删除表中全部数据
     DELETE FROM TABLE_NAME WHERE

四.SQL数据检索（*）

（1）数据的检索 -- select

select  *  from  这里注意在数据检索中，如果能不用*就不用，尽量用所需要的字段名进行检索 就像下面

      SELECT  FName  as  姓名 ,  Fage  as 年龄  , Fsalary  as  月薪 FROM  T_Employee

      < as 的作用  : 取别名 >

 

 select COUNT(*) FROM Person1
 select MAX(Age) FROM Person1
 select MIN(Age) FROM Person1
 select avg(Age) FROM Person1
 select sum(Age) FROM Person1
 select COUNT(*) FROM Person1 Where

（2）数据的排序 -- order by

 

 select * from Person1 order by Age (ASC升序 DESC降序) ascend   descend
 select * from Person1 order by Age DESC,Number ASC;按照age降序,number升序

 order by 要放在where句子后面

（3）模糊搜索 -- like

通配符
     单字符通配符“ _”,它匹配单个出现的字符, LIKE '_erry'    任意第一个字符
     多字符通配符“ %”,它匹配任意长度字符,0到多个 LIKE '%n%'   包含n的所有字符
（4）sql中的null的理解

null相当于不知道
     select * from Person1
     where Name  is null        或者    where Name is not null
（5）多值匹配
     select * from Person1
     where Age=13 or Age=15 or Age=30
     使用where Age  in(13,15,30) 简单
     where Age>13 and Age <15  清晰
     (where Age between 13 and 15) 比较少用
（6）数据分组
     Group by  放在 where 后面
     select Age as 年龄,count(*) as 人数 from Person1
      group by Age
     根据年龄进行分组,

（7）Having语句
     聚合函数不应出现在where中
     having出现在group by之后
     select Age as 年龄,count(*) as 人数 from Person1
     group by Age
      Having count(*) > 1
     order by 人数 ASC
     注意先后顺序
     Having是对分组后信息的过滤,能用的列和select中能用的列是一样
（8）Top  限制结果集的行数
     select  top 3 * from ...  前三行
     select top 3 * from Person1
     where Number Not In (select top 5 Number from Person1 order by Age DESC)
     order by Age DESC
     显示第6条以下的前3条行
     SQL2005后增加了Row_Number函数简化实现,后面会讲
(9) Distincct  去掉重复数据
     select  DISTINCT Age as 年龄 from Person1
     select Age,NickName from Person1
     select DISTINCT Age,NickName from Person1
     DISTINCT针对整行重复,不是一个字段的重复, 这里的指的是你检索出来的行,如果Age,NickName两个字段,就是这样的两个字段的行,其实实际也很符合,如果nickname不一样,不显示就失去数据了

(10)union 联合

     select  Age  from  Person1

      union

      select  Age  from  Person2

      组合成一列  , 关键字段必须一致 , 数据类型要相容

      select  Age ,  NickName  from  Person1

      union

      select  Age ,  ' 没有昵称 '  from  Person2

     会去掉完全重复的数据,union all 就可以不去掉(一般都用union all)

案例1

 select '正式工最高年龄',MAX(Age) FROM Person1
 union all
 select '正式工最低年龄',MAX(Age) FROM Person1
 union all
 select '临时工最高年龄',MAX(Age) FROM Person2
 union all
 select '临时工最高年龄',MAX(Age) FROM Person2

案例2
 查询正式工的信息,包括工号,工资,最后一行加上所有员工的工资合计
 select FNumber,FSalary FROM T_Employee
 union all
 select '工资合计',sum(Fsalary)

(11)数字函数
     ABS():求绝对值
     CEILING():舍入到最大整数  3.33->4, -3.6->-4
     FlOOR() :舍入到最小整数
     ROUND(): 四舍五入 有两个参数 ROUND(半径,小数精度)
(12)字符串函数
     LEN() : 求字符串长度 LEN('abc') : 3  ??字符串右侧的空格SQL中似乎不识别,比如len('abd   ')=3
     LTRIM():字符串左侧的空格去掉
     RLTRIM() :字符串右侧的空格去掉
     SUBSTRING()  : 取子字符串, 三个参数, SUBSTRING(主字符串,起始,长度),从1开始计数,substring('123',1,1)结果是'1'
(13)日期函数
     GETDATE() : 取得当前时间
     DATEADD(datepart,number,date)     : 计算增加以后的日期number是日期增量
               quarter   qq,q  季度
               month     mm,m  月份
               dayofyear dy,y  当年度的第几天
               day                dd,d  日
               week               wk,ww 当年度的第几周
               weekday   dw,w  星期几
               hour      hh          小时
               minute          mi,n     分
               second          ss,s     秒
               millisecond     ms     毫秒 
                    select getdate() as 今天
                    ,DateAdd(day,-1,getdate()) as 前天
                    ,DateAdd(hh,-1,getdate()) as 前一个小时
                    ,DateAdd(mm,-1,getdate()) as 前一个月

  例1:select DateDiff(hh,getdate(),DateAdd(hh,1,getdate())) 结果为1
  例2:select DateDiff(year,FinDate,getdate()),count(*) as 入职年数 from T_Employee
     group by DateDiff(year,FinDate,getdate())
     按照入职的年数进行分组
 DatePart(datepart,date) : 返回一个日期的特定部分
  例1:select DatePart(year,getdate()),DatePart(month,getdate())
  例2:select DatePart(year,FInDate),count(*) from T_Employee
    group by DatePart(year,FInDate)

(14)类型转换函数   

 CAST(expression AS data_type)
 CONVERT(data_type,expression)
 select cast('123' as int )+1,cast('2008-09-09' as datetime),
 convert(int,'333'),convert(datetime,'2008-09-09')
 select DatePart(year, cast('2008-09-09' as datetime)) : 确保类型正确

(15) Isnull  空值控制函数

 ISNULL(expression,value): 如果 expression不为空,返回expression,否则返回value
  例子:select ISNULL(Name,'佚名') from Person1

 

(16) case函数用法1   面试中用到很多
     Case i
     when value1 then returnvalue1
     when value2 then returnvalue2
     ...
     例子1:
          

 select FName,
 (
 Case FLevel
 when 1 then '普通客户 '
 when 2 then '会员 '
 when 3 then 'VIP'
 else ' 未知类型'
 end
 ) as 客户类型
 from T_Customer

  
例子2 :           

 select FName,
 (
 case
 when Fsalary< 2000 then '低收入'
 when Fsalary>= 2000 and Fsalary <5000 then '中等收入 '
 when Fsalary>= 5000 then '高收入'
 end
 ) as 收入水平
 from T_Customer

五.课后练习（＊）

练习1

 select 单号 ,
 (
 case
 when 金额 >0 then 金额
 else 0
 end
 ) as 收入,
 (
 case
 when 金额 <0 then ABS( 金额)
 else 0
 end
 ) as 支出
 from T_1

练习二
2008/8/8     拜仁     胜
2008/8/9     奇才     胜
2008/8/9     湖人     胜
2008/8/10     拜仁     负
2008/8/8     拜仁     负
2008/8/12     奇才     胜

显示成
拜仁     1     2
湖人     1     0
奇才     2     0

下面是SQL语句:

 CREATE TABLE [T_Scores](
      [Date] [datetime] NULL,
      [Name] [nvarchar] (50) NULL,
      [Score] [nvarchar] (50) NULL
 );
 INSERT [T_Scores] ( [Date],[Name] ,[Score]) VALUES(CAST ('2008-08-08' as datetime),N '拜仁' ,N' 胜')
 INSERT [T_Scores] ( [Date],[Name] ,[Score]) VALUES(CAST ('2008-08-09' as datetime),N '奇才' ,N' 胜')
 INSERT [T_Scores] ( [Date],[Name] ,[Score]) VALUES(CAST ('2008-08-09' as datetime),N '湖人' ,N' 胜')
 INSERT [T_Scores] ( [Date],[Name] ,[Score]) VALUES(CAST ('2008-08-10' as datetime),N '拜仁' ,N' 负')
 INSERT [T_Scores] ( [Date],[Name] ,[Score]) VALUES(CAST ('2008-08-08' as datetime),N '拜仁' ,N' 负')
 INSERT [T_Scores] ( [Date],[Name] ,[Score]) VALUES(CAST ('2008-08-12' as datetime),N '奇才' ,N' 胜')
 SELECT Name ,
 SUM(
 CASE Score
 WHEN ' 胜' THEN 1
 ELSE 0
 END
 ) AS 胜,
 SUM(
 CASE Score
 WHEN ' 负' THEN 1
 ELSE 0
 END
 ) AS 负
 FROM T_Scores
 GROUP BY Name

练习3
     创建一张表,记录电话呼叫员的工作流水,记录呼叫员的编号,对方号码,通话开始时间,通话结束时间.
建表,插数据都自己写SQL语句
     要求:输出所有数据中通话时间最长的5条记录.
               输出所有数据中拨打长途号码(对方号码以0开头)的总时长
               输出本月通话总时长最多的前三个呼叫员的编号
               输出本月拨打号码次数最多的前三个呼叫员的编号
               输出所有数据的拨号流水,并且在最后的一行添加总呼叫次数
                    -记录呼叫员编号,对方号码,通话时长
                    -...
                    -汇总[市内号码总时长][长途号码总时长]
     SQL的代码:

 CREATE TABLE [T_PhoneWork] (
      [id] int identity (1, 1) primary key ,
      [WorkNum] [nvarchar] (50) NOT NULL,
      [PhoneNum] [nvarchar] (20) NULL,
      [StartTime] [datetime] NULL,
      [EndTime] [dateTime] NULL
 );
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100001' ,'13888888888', CAST('2013-1-1 7:10:10' AS datetime),CAST ('2013-1-1 7:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100002' ,'13887888788', CAST('2013-1-1 8:10:10' AS datetime),CAST ('2013-1-1 8:30:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100003' ,'13888688868', CAST('2013-1-1 7:20:10' AS datetime),CAST ('2013-1-1 8:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100004' ,'13838883888', CAST('2013-1-1 8:10:10' AS datetime),CAST ('2013-1-1 10:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100001' ,'13844888888', CAST('2013-1-1 10:10:10' AS datetime),CAST ('2013-1-1 12:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100002' ,'013888885888', CAST('2013-1-1 13:16:10' AS datetime),CAST ('2013-1-1 14:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100003' ,'13888588888', CAST('2013-1-1 14:10:10' AS datetime),CAST ('2013-1-1 16:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100001' ,'013868886888', CAST('2013-1-1 12:10:10' AS datetime),CAST ('2013-1-1 17:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100004' ,'13788878877', CAST('2013-1-1 14:10:10' AS datetime),CAST ('2013-1-1 16:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100001' ,'13787878877', CAST('2013-1-1 16:10:10' AS datetime),CAST ('2013-1-1 18:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100002' ,'13787878877', CAST('2013-1-1 12:10:10' AS datetime),CAST ('2013-1-1 14:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100003' ,'13788078877', CAST('2013-1-1 16:10:10' AS datetime),CAST ('2013-1-1 17:20:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('100005' ,'13799078877', CAST('2013-5-1 12:10:10' AS datetime),CAST ('2013-5-2 22:20:10' AS datetime))
 CREATE TABLE [T_PhoneWork] (
      [id] int identity (1, 1) primary key ,
      [WorkNum] [nvarchar] (50) NOT NULL,
      [PhoneNum] [nvarchar] (20) NULL,
      [StartTime] [datetime] NULL,
      [EndTime] [dateTime] NULL
 );
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('001' ,'020888888', CAST('2010-7-10 10:00:00' AS datetime),CAST ('2010-7-10 10:05:03' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('001' ,'020888888', CAST('2010-7-11 13:00:00' AS datetime),CAST ('2010-7-11 13:01:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('001' ,'8934343434', CAST('2010-7-11 14:06:00' AS datetime),CAST ('2010-7-11 14:09:00' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('002' ,'9837434777', CAST('2010-7-13 21:06:00' AS datetime),CAST ('2010-7-13 21:08:08' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('002' ,'0213434343', CAST('2010-6-29 20:11:00' AS datetime),CAST ('2010-6-29 20:16:06' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('001' ,'787878778', CAST('2010-7-15 13:16:00' AS datetime),CAST ('2010-7-15 13:26:00' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('003' ,'0334343444', CAST('2010-7-13 11:16:00' AS datetime),CAST ('2010-7-13 11:17:09' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('003' ,'676777333', CAST('2010-7-19 19:26:02' AS datetime),CAST ('2010-7-19 19:30:33' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('001' ,'89923434333', CAST('2010-6-19 15:16:02' AS datetime),CAST ('2010-6-19 15:26:10' AS datetime))
 INSERT INTO [T_PhoneWork]([WorkNum], [PhoneNum],[StartTime] ,[EndTime]) VALUES('004' ,'400400400', CAST('2010-6-19 15:16:02' AS datetime),CAST ('2010-6-19 15:26:10' AS datetime))

1.输出所有数据中通话时间最长的5条记录.

 SELECT TOP 5 *FROM T_PhoneWork
 ORDER BY (DateDiff (s, StartTime,EndTime )) DESC

2.输出所有数据中拨打长途号码(对方号码以0开头)的总时长

 SELECT SUM (
 DateDiff(s ,StartTime, EndTime)
 ) AS 长途总时长
 FROM T_PhoneWork
 WHERE PhoneNum LIKE '0%'

3.输出本月通话总时长最多的前三个呼叫员的编号

 SELECT TOP 3 WorkNum,count (*) FROM T_PhoneWork
 WHERE DateDiff (month, StartTime,getdate ()) = 0
 group by WorkNum
 order by sum( datediff(s ,StartTime, EndTime)) DESC

 SELECT TOP 3 WorkNum FROM T_PhoneWork
 WHERE DateDiff (month, StartTime,'2010-7-2 9:12:00' ) = 0
 group by WorkNum
 order by sum( datediff(s ,StartTime, EndTime)) DESC

4.输出本月拨打号码次数最多的前三个呼叫员的编号

 select top 3 WorkNum from T_PhoneWork
 WHERE DateDiff (month, StartTime,getdate ()) = 0
 group by WorkNum
 order by count(*) DESC

5.注意,union的两个原则,类型要相容,这个容易犯错,谨记
输出所有数据的拨号流水,并且在最后的一行添加总呼叫次数
                    -记录呼叫员编号,对方号码,通话时长
                    -...
                    -汇总[市内号码总时长][长途号码总时长]
                   

 select WorkNum as 呼叫员编号,PhoneNum as 对方号码 ,DateDiff( s,StartTime ,EndTime) as 通话时长
 from T_PhoneWork
 union all
 select ' 汇总',
 cast(
 sum(
 CASE substring (PhoneNum, 1,1 )
 when '0' then 0
 else DateDiff (s, StartTime,EndTime )
 end
 ) as nvarchar( 50)),
 cast(
 sum(
 CASE substring (PhoneNum, 1,1 )
 when '0' then DateDiff(s ,StartTime, EndTime)
 else 0
 end
 ) as nvarchar( 50))
 from T_PhoneWork

老师做的

 select WorkNum as 呼叫员编号,PhoneNum as 对方号码 ,DateDiff( s,StartTime ,EndTime) as 通话时长
 from T_PhoneWork
 union all
 select ' 汇总',
 cast(
 sum(
 CASE
 when PhoneNum not like '0%' then DateDiff( s,StartTime ,EndTime)
 else 0
 end
 ) as nvarchar( 50)),
 cast(
 sum(
 CASE
 when PhoneNum like '0%' then DateDiff (s, StartTime,EndTime )
 else 0
 end
 ) as nvarchar( 50))
 from T_PhoneWork

30.索引
     全表扫描:对数据进行检索(select)效率最差的是全表扫描,就是一条条的找
     如果没有目录,查汉语字典就要一页页的翻,而有了目录只要查目录即可.为了提高检索速度,可以为经常进行检索的列添加索引,相当于创建目录
     建立索引优点:检索速度快
                         缺点:占磁盘空间,降低编辑表的操作,如insert,update
     (*)即使创建了索引,仍然有可能全表扫描,比如like,函数,类型转换等

31.表连接jion   主流不增加外键
     有客户表 和 订单表 ,关联查询   [table1] as [别名] join  [table2]  as [别名] on [约束条件]

     例子:显示所有年龄大于25的客户订单号,客户名字,客户年龄
            select  o .  BillNo , p  . Name ,  p . Age

       from  T_Order  as  o  join  Person1  as  p  on  o . CustomerId  = p .  Number

        where  p .  Age  >  25

     例子2:大于平均年龄

select  o .  BillNo , p  . Name ,  p . Age

from  T_Order  as  o  join  Person1  as  p  on  o . CustomerId  = p .  Number

where  p .  Age  >  ( select  avg ( Age  )  from  Person1 )

    
32.子查询
     将一个查询语句作为一个结果,作为一个结果集供其他SQL语句使用,就像使用普通的表一样,被当做结果集的查询语句被称为子查询.
     左右可以使用表的地方几乎都可以使用子查询来代替
     select * from
     (select * from Person1) as p1
     单值作为子查询  一行一列,单值查询
     单列多行的数据可看做一个数据集,单列多行的子查询
     例子

       

 select * from T_Reader
    where FYearOfJoin IN( 2001,2003 )
    select * from T_Reader
    where FYearOfJoin IN
   (select FYearPublished from T_Book) --查询书记出版那年加入协会的人的信息

      限制结果集  .  返回第行到第行的数据

 select * from
 (
 select ROW_NUMBER() OVER( ORDER BY Age ASC) AS rownum ,
 Number ,Name, Age,NickName From Person1
 ) as e1
 where e1. rownum>=3 and e1. rownum<=5

     经典的应用是: 分页的操作

     select * from (
 select ROW_NUMBER()
 over(order by k1 .TimeDiff DESC ) as rownum,k1 .WorkNum, k1.TimeDiff as d
 from (
 select WorkNum, sum(datediff (second, StartTime,EndTime )) as TimeDiff
 from T_PhoneWorkTest
 group by WorkNum
 )as k1
 ) as e1
 where e1. rownum >= 2 and e1. rownum <=3